Ultrafast Protein Conformational Alterations in ... - ACS Publications

Department of Physics, University of Kaiserslautern, D-67663 Kaiserslautern, Germany, and Department of Organic Chemistry, Weizmann Institute of Scien...
1 downloads 0 Views 3MB Size
J. Phys. Chem. B 2009, 113, 7851–7860

7851

Ultrafast Protein Conformational Alterations in Bacteriorhodopsin and Its Locked Analogue BR5.12 Ruth Gross,† Christian Schumann,†,‡ Matthias M. N. Wolf,† Johannes Herbst,†,§ Rolf Diller,*,† Noga Friedman,| and Mordechai Sheves| Department of Physics, UniVersity of Kaiserslautern, D-67663 Kaiserslautern, Germany, and Department of Organic Chemistry, Weizmann Institute of Science, RehoVot, Israel ReceiVed: NoVember 14, 2008; ReVised Manuscript ReceiVed: April 3, 2009

Bacteriorhodopsin, reconstituted with a sterically “locked” retinal chromophore, BR5.12, has frequently been used to elucidate elementary photoinduced processes in the native pigment bacteriorhodopsin. In this work, the vibrational response of BR5.12 to photoexcitation is investigated by means of femtosecond time-resolved mid-infrared and UV-vis spectroscopy. The electronically excited state of BR5.12 decays with a time constant of 18 ps. Neither in the UV-vis nor in the mid-IR spectral region are indications found for chromophore photoproducts, besides the full recovery of the electronic ground state. However, vibrational bands are observed at around 1660 and 1550 cm-1 in the protein amide I and amide II band regions, respectively. They are formed within a few picoseconds or even instantaneously. Thus, they appear faster than the S1 decay and persist for at least 130 ps, i.e., for much longer than the S1 lifetime. These findings strongly suggest that the observed bands must be assigned to protein vibrations and that they are not caused by a photoinduced temperature rise. Thus, for the first time, ultrafast protein vibrational changes are detected in BR5.12, that are not associated with isomerization. Possibly they can be related to the enhanced chemical reactivity of photoactivated BR5.12 reported in the literature. In wild-type bacteriorhodopsin, bands with very similar spectral and kinetic characteristics are observed, suggesting that they might originate from a similar mechanism which is not isomerization. A plausible mechanism is a polarization induced protein conformational change, as discussed in the literature. Introduction Bacteriorhodopsin (BR) is a retinal protein acting as a light driven proton pump in halobacteria and belongs together with the chloride pump halorhodopsin (HR) and the light sensors sensory rhodopsin I and II (SRI/II) to the family of archaebacterial rhodopsins. It consists of seven transmembrane R-helices enclosing an all-trans retinal molecule as a functional group which is covalently bound to lysine 216 via a protonated Schiff base. Retinal proteins undergo a light induced all-trans to 13cis double bond isomerization of the chromophore on a subpicosecond to picosecond time scale as the first step of a photocycle which is completed within milliseconds via a number of thermally driven reaction steps. The primary chromophore dynamics of retinal proteins have been studied extensively, e.g., by ultrafast UV-vis, resonance Raman, and mid-infrared absorption spectroscopy (for a recent review, see ref 1 and work cited therein2-5). In a simplified scheme of the primary reaction of BR, the optically excited Franck-Condon state is depopulated on the femtosecond time scale, mainly along C-C stretching modes. The transition from the excited electronic state to the electronic ground state proceeds from state I to state J which then decays on the time scale of a few picoseconds to the longlived (microseconds) K intermediate. Isomerization has occurred * Corresponding author. E-mail: [email protected]. Phone: +49631-205-2323. Fax: +49-631-205-3902. † University of Kaiserslautern. ‡ Present address: Leibniz Institute for New Materials, D-66123 Saarbrücken, Germany. § Present address: Fraunhofer Institute for Physical Measurement Techniques, Heidenhofstr. 8, D-79110 Freiburg, Germany. | Weizmann Institute of Science.

in the electronic ground state J, as revealed from vibrational resonance Raman and infrared analysis.5-7 Within this model, the J-K transition is believed to comprise vibrational relaxation (e.g., vibrational cooling) and conformational changes accompanying or following chromophore isomerization. In contrast to the primary photoinduced dynamics of the retinal chromophore, very little is known about how and when the primary perturbation, comprising chromophore excitation and reaction, couples to the chromophore binding pocket as a prelude for following protein dynamics that execute the specific biological function. Vibrational bands of the protein moiety and of protein bound water molecules have been observed during the earliest part of the photocycle in time-resolved Fourier transform infrared (FTIR), ultrafast transient IR, and low temperature IR measurements on BR,5,8-11 HR,12,13 SRII,14-16 and proteorhodopsin (PR)17-19 around 1550 and 1660 cm-1. Kandori et al. reported changes of O-D stretching vibrations of three water molecules belonging to a pentagonal cluster in the Schiff base region in the K state of BR.20,21 Similar features of water bands were observed in SRII, indicating analogous structural changes of internal water molecules in both systems.22 This is supported by X-ray studies on BR and SRII demonstrating a rearrangement of specific amino acid residues and a water cluster in the vicinity of the retinal Schiff base in the K state.23,24 Whereas the cited protein conformational changes in the wildtype retinal proteins can be rationalized in the framework of the isomerising chromophore, i.e., a strong steric perturbation, a broad variety of experiments employing artificial pigments with nonisomerizable chromophores have dealt with processes induced merely by electronic excitation of the respective

10.1021/jp810042f CCC: $40.75  2009 American Chemical Society Published on Web 05/07/2009

7852

J. Phys. Chem. B, Vol. 113, No. 22, 2009

Figure 1. Chemical structure of the “locked” prosthetic group which replaces retinal in BR5.12.

chromophore.25 Thereby, the evolution of the electronically excited state has been studied and compared to the native chromophore.26 Moreover, various other photoinduced processes pertaining to the protein moiety have been observed. They include conformational change sensed by atomic force microscopy,27 cleavage of the CdNH+ bond via hydroxylamine reactions28 and reduction-oxidation reactions of spin labels covalently bound to appropriate BR mutants.29 The underlying mechanism suggested for the observed photoinduced protein alterations in “locked” retinal proteins employs the change of the electronic dipole moment of the chromophore upon photoexcitation which polarizes the protein matrix30-32 and may induce long living protein conformational changes. It is further suggested that, in wild-type systems, a similar primary protein polarization takes place and is coupled to the chromphore isomerization in an essential way. To further explore the light induced processes in BR pigments with nonisomerizable chromophores, we studied in this work the ultrafast response of the artificial BR pigment, BR5.12, in the mid-IR after photoexcitation. BR5.12 is derived from a retinal analogue, sterically fixed around the C13dC14 double bond with a five-membered carbon ring (Figure 1), preventing the all-trans to 13-cis isomerization. Ultrafast visible transient absorption experiments showed that the electronic excited state decays with a time constant of ca. 17 ps, without isomerization or a photocycle, back to the electronic ground state.25,28 Coherent anti-Stokes Raman spectroscopy (CARS) measurements33 have indicated that following excitation to the Franck-Condon state the chromophore relaxes to the T5.12 state on the S1-surface which decays to the BR5.12 ground state as the only reaction pathway. Here, we present ultrafast transient IR experiments on BR5.12 and BR. They show that in both BR5.12 and BR absorbance differences appear in the amide I and amide II regions within less than 5 ps or even instantaneously after photoexcitation and persist longer than the experimental time scale (130 ps), in particular for much longer than the respective excited electronic state lifetime. The observation that the amide I and amide II features in both BR5.12 and BR bear almost identical spectral and temporal characteristics suggests the existence of a mechanism distinct from isomerization as a cause for a primary protein response in photoexcited retinal proteins. Materials and Methods Sample Preparation. Suspensions of bacteriorhodopsin were prepared according to published procedures.34 Apo-membrane was prepared by illumination of bacteriorhodopsin in the presence of hydroxylamine (2 M, pH 7).35 The synthetic “locked” retinal analogue, retinal-5.12 (five-membered ring spanning carbons 12,14), was prepared as previously described.36 BR5.12 was prepared by incubating apo-membrane (100 mM NaCl, 50 mM phosphate buffer, pH 7) with retinal-5.12 (1.5 equiv, in EtOH, 49:1 v/v) in the dark, at room temperature, for 15 days. The pigment was washed with water and concentrated to 0.5 mM suspension (15 mM KCl, 1 mM phosphate buffer, pH 7). Films of BR and BR5.12 were prepared on a CaF2 window of 1.5 in. diameter and sealed by a second window. The samples

Gross et al. were hydrated just sufficiently to ensure full protein function of BR (checked via determination of the duration of the photocycle) and to avoid unnecessary absorption losses by water. The optical density in the visible absorption maximum (λmax,BR ) 570 nm and λmax,BR5.12 ) 580 nm) was ca. 1 OD. During all measurements, the sample was rotated and moved laterally normal to the laser beam direction in order to ensure fresh sample conditions at each laser pulse. Sample integrity was confirmed by steady state FTIR and UV-vis spectra before and after the measurements. All experiments were performed at room temperature. Transient Absorption Spectroscopy. A commercial titanium-sapphire laser system (CPA 2001, Clark-MXR Inc., Dexter, MI) with a repetition rate of 635 Hz was used as a source for generating ultrashort pump and probe pulses by means of nonlinear optics as described earlier.13 For the experiments in the visible regime (VIS/VIS), two home-built noncollinear parametric amplifiers (NOPA), tunable between 470 and 765 nm, were employed as the pump and probe sources. After prism compression, the pulse duration was determined to be 50-70 fs fwhm (full width at half-maximum). The experimental time zero and the system response were obtained by transient absorption measurements on a rhodamin101 solution in ethanol. After transmission through the sample, the probe pulses passed through a monochromator and the intensity at the selected wavelength was detected by a photodiode. For experiments in the mid-IR (VIS/IR), the excitation pulses in the visible were obtained as described above, with a focal width of about 200 µm and a pulse energy of 0.4 µJ at the sample position. The IR pulses were generated by a two-stage optical parametric amplifier (OPA) followed by difference frequency generation. The transmitted IR beam with a spectral width of 100 cm-1 was dispersed in a polychromator, and the spectrally resolved intensity was detected by a 32-channel mercury-cadmium-telluride array (Infrared Systems). For earlier experiments on BR, a detector array with only 10 elements was used. The system response (250-330 fs fwhm) and time zero were determined by pump-probe experiments on a thin silicon sample. Both in the visible and in the IR measurements, every second pump pulse was chopped and the pump induced absorption differences ∆A(t,λpr) were measured as a function of the delay time t between pump and probe pulse at probe wavelength λpr and were evaluated on a single-shot basis. Negative absorption changes indicate the disappearance of IR absorption, e.g., photoinduced depopulation of electronic ground state vibrations (bleach bands). Positive absorbance changes display the absorption of newly populated states. The detected absorbance changes were quantitatively analyzed by a global multiexponential fit (eq 1) for delay times >300 fs using N

∆A(t, λpr) ) A0(λpr) +

∑ Ai(λpr) · e-t/τ

i

(1)

i)1

with A0 being the pump induced difference absorption spectra after long delay times (up to 150 ps in our experiments) and Ai(λpr) being the decay associated spectra (DAS) of the corresponding time constants τi. Results VIS/VIS Measurements. Since the transient IR experiments were performed on hydrated films instead of aqueous suspensions, it was necessary to first characterize the photoinduced electronic state dynamics of the artificial pigment BR5.12 in

Ultrafast Protein Alterations in BR and BR5.12

Figure 2. Absorption transients of BR and BR5.12 after excitation at 570 and 580 nm, respectively, at a probe wavelength of 480 nm. The inset shows the same data on a shorter time scale.

this sample preparation. Therefore, transient visible absorption experiments were performed on equivalent films of BR5.12 and BR and compared to results from suspensions. The samples were excited at the respective absorption maximum, 570 nm (BR) and 580 nm (BR5.12), and probed at 480 nm. The observed instantaneous positive absorbance difference signal at 480 nm (Figure 2) is assigned to excited state absorption for both BR and BR5.12.26 A monoexponential fit of the BR transient yields a time constant of 0.5 ( 0.1 ps. After the S1 decay, the difference signal becomes negative and stationary on the time scale of the experiment. It is assigned to the residual bleach (at 480 nm) of the electronic ground state absorption band of BR570. At early times, the initial BR570 ground state bleach is superimposed by the much stronger S1 absorption at this wavelength. Since the isomerization quantum yield is unequal to zero (0.6437-39), only a fraction of the initial bleach recovers upon S1 decay and leads to the residual negative signal. The difference signal of BR5.12 is also positive at 480 nm but decays much slower with very small negative amplitude at long delay times. A biexponential fit yields time constants of 0.7 ( 0.1 and 17.7 ( 0.3 ps. The long time constant describes the S1 decay of BR5.12, whereas the short time constant is attributed to a small amount of BR in the BR5.12 sample. Both excited state lifetimes are in accordance with previous measurements in suspensions2-4,40 and show that the electronic state dynamics are not changed in the film preparation. Assuming almost identical extinction coefficients for the electronic ground states40 of BR5.12 and BR, as well as for the respective excited states, and taking into account the spectral width of the exciting laser pulse (30 nm), the amount of residual BR in the BR5.12 sample was determined to be 15-20%. It was derived first from the ratio of the maximum positive difference signal to the negative difference signal at long delay times at 130 and 40 ps in BR5.12 and BR, respectively, and second from the amplitude ratio of the two components in the biexponential fit of the BR5.12 transient. VIS/IR Measurements. Fingerprint Region. The transient IR experiments on BR5.12 in the spectral region from 1175 to 1215 cm-1 further characterize the sample and its vibrational dynamics. In this spectral region, C-C stretching modes coupled to C-H bending modes serve as marker bands for the chromophore configuration. The observed IR signals provide evidence that the chromophore in BR5.12 does not isomerize upon photoexcitation and fully recovers to the initial ground state (all-trans configuration).

J. Phys. Chem. B, Vol. 113, No. 22, 2009 7853

Figure 3. IR difference spectra of BR5.12 in the fingerprint region after excitation at 580 nm at various delay times.

Transient IR difference spectra of BR5.12 (Figure 3) show early bleach at around 1202 and 1178 cm-1 (t ) 0.15 ps). The late spectrum at t ) 130 ps exhibits a small negative band around 1202 cm-1 and a broader positive band centered around 1190 cm-1. The bands at 1202 and 1190 cm-1 are assigned to C-C stretching vibrations of the all-trans BR ground state BR57041 and the 13-cis K state,6,7 respectively. The strong negative band at 1178 cm-1 can be assigned to a C-C vibration of the BR5.12 ground state in accordance with CARS measurements33 where the band is observed at 1182 cm-1. There, this band shifts only very little upon T5.12 formation, i.e., to 1186 cm-1. Since in the IR difference spectra this band remains negative throughout the entire excited state lifetime, it can be concluded that its IR absorption cross section decreases significantly upon electronic excitation. The global analysis of this spectral region (Figure 4a) yields three decay associated spectra (DAS) with time constants of 0.5 ( 0.1, 4.7 ( 0.2, and 18.5 ( 0.2 ps and a spectrum at “infinite” delay time (A0) which is nearly the same as the difference spectrum at 130 ps in Figure 3. The 0.5 and 4.7 ps DAS describe the evolution of the residual amount of BR in the sample and are nearly identical with those obtained in earlier experiments on BR (for experimental details, see ref 5) (Figure 4b) besides small variations of the observed time constants. Here, the 0.5 ps DAS represents the formation of the C-C stretching vibration at 1190 cm-1 (J state). This band is characteristic for the 13-cis configuration6,7 and signals the chromophore isomerization with 0.5 ps5 time constant. The 4.7 ps DAS with a negative lobe at 1200 cm-1 and a positive lobe below 1197 cm-1 indicates a small blue shift of the 13-cis vibration around 1190 cm-1, in accordance with vibrational cooling and torsional relaxation of the BR chromophore. Consequently, the 18.5 ps DAS with a strong negative band around 1178 cm-1 and weak positive bands at 1198 and 1208 cm-1 (vibrations of the electronically excited state) reflects the vibrational changes of the BR5.12 chromophore during its excited state lifetime, about 18 ps, as found in the VIS/VIS measurements. Thus, the difference spectrum at long delay times, i.e., after K formation of the BR fraction and after electronic ground state recovery of the BR5.12 fraction, can fully be attributed to the K-BR difference spectrum of the BR fraction of the sample. Note that at 1178 cm-1 no remaining bleach is observed at long delay times. Instead, the small positive absorption difference at 130 ps is caused by the absorption of the K state. This strongly

7854

J. Phys. Chem. B, Vol. 113, No. 22, 2009

Gross et al.

Figure 5. IR difference spectra of BR in the amide I and amide II regions after excitation at 570 nm at selected delay times.

Figure 4. Decay associated spectra (DAS) of the fingerprint region of BR5.12 (a) and of a BR sample (b). The DAS of the BR sample (b) were obtained in earlier experiments on BR (small differences in band position with respect to part a are due to a slightly different wavenumber calibration).

suggests and confirms earlier reports25 that BR5.12 does not undergo photoinduced isomerization and completely recovers to its initial ground state. Amide I and Amide II Regions. In this paragraph, we compare results of BR (pure sample) and BR5.12 in the spectral region 1680-1500 cm-1. Here, not only chromophore bands but also vibrations of the protein and of protein bound water molecules are expected, especially, H2O bending and peptide CdO stretching modes (amide I) around 1660 cm-1 and peptide CsN stretching modes (amide II) around 1550 cm-1. The results reveal photoinduced IR difference signals in BR5.12 in the amide I and amide II regions that appear fast, i.e., within less than 5 ps, and persist much longer than its excited state lifetime. Bands with similar spectral and kinetic features are observed in BR. BR. The IR absorbance changes in BR confirm the biphasic chromophore dynamics during its all-trans to 13-cis isomerization as reported earlier.5 Moreover, absorbance difference bands in the amide I and amide II regions are observed associated with a unique time constant of 11 ps. The bleach signals at 1529 and 1640 cm-1 observed in BR absorbance difference spectra (Figure 5) are assigned to the ethylenic CdC and the CdNH stretching vibration, respectively, of the chromophore in its all-trans configuration.41 The positive

Figure 6. IR difference spectra of BR at 40 ps and of BR5.12 at 130 ps in the amide II region at long delay times. In order to improve the statistics, at each wavenumber, all absorbance data in the stationary time regime, i.e., between 35 and 60 ps (BR) and between 60 and 150 ps (BR5.12), were averaged to obtain the depicted difference spectra and errors. The circle marks the spectral region (1545-1560 cm-1) of observed protein associated bands.

(product) signals around 1516, 1610, and 1622 cm-1 correspond to the K state.8 The instantaneous positive absorption difference at 1570 cm-1 decays in parallel with the excited electronic state (see below) and is thus assigned to an S1 vibrational mode of the chromophore. Absorbance changes observed around 1550 (1545-1560 cm-1) and above 1650 cm-1 (1655-1670 cm-1) (Figures 5, 6, and 7) are suggested to be nonchromophore bands for the following reasons. Above 1640 cm-1, the position of the CdNH stretch, no further chromophore modes are expected41,42 (not taking into account CsH and NsH stretching vibrations and assuming a sufficiently homogeneous distribution of the BR ground state). Similar bands, in shape and spectral position, have been observed earlier in many IR experiments concerning the primary reaction in retinal proteins.8,43,44 Concerning the band around 1550 cm-1, a number of studies have proposed13,14,19 or have provided16,17 evidence for the existence of an early protein response in BR, HR, PR, and SRII in this spectral region based on femtosecond IR experiments on isotopically labeled proteins. A global analysis of the BR data according to eq 1 yields three time constants: τ1,BR ) 0.6 ( 0.1 ps, τ2,BR ) 3.2 ( 0.1 ps,

Ultrafast Protein Alterations in BR and BR5.12

Figure 7. IR difference spectra of BR at 40 ps and of BR5.12 at 130 ps in the amide I region at long delay times. (For data and error handling, see Figure 6.) The circle marks the spectral region (1655-1670 cm-1) of observed protein associated bands.

Figure 8. DAS as a result of the global analysis in the amide I and amide II regions of BR.

and τ3,BR ) 11.1 ( 0.6 ps. The corresponding DAS are shown in Figure 8. They are in accordance with a branching reaction from the excited electronic state to the isomerized electronic ground state J and to the initial all-trans state with 0.6 ps, followed by vibrational cooling and torsional relaxation with about 3 ps in both paths (K formation and BR570 recovery), as described earlier for BR5 and SRII.14 In the ethylenic stretch region, the negative amplitude of the τ1,BR DAS around 1511 cm-1 indicates the fast formation of the J state (early band at 1511 cm-1 in Figure 5), which decays to K (1516 cm-1)7 with τ2,BR (positive amplitude at 1511 cm-1 of the τ2,BR DAS). The appearance of the red-shifted ethylenic stretch follows the linear correlation45,46 between the ethylenic stretch wavenumber and the electronic absorption maximum (λmax) of the J state. The negative amplitudes of the τ1,BR DAS and the τ2,BR DAS above 1520 cm-1 display the fast formation of hot BR570 and its (blueshifted) vibrational ground state, respectively. The DAS in the region of the CdNH stretching vibration around 1640 cm-1 exhibits analogous spectral features and can be interpreted accordingly. Moreover, in addition to earlier studies,5 the 11 ps DAS is clearly observed. Its amplitude spectrum exhibits a positive peak at 1522 cm-1 and negative peaks at 1555, 1620, and around

J. Phys. Chem. B, Vol. 113, No. 22, 2009 7855

Figure 9. Difference spectra in the amide I and amide II regions of BR5.12 at selected delay times.

1660 cm-1 (small). The minima at 1555 and 1660 cm-1 correspond well in their spectral position with those features described as nonchromophore bands above. Consequently, the 11 ps time constant suggests a process associated primarily with the protein moiety and not with the chromophore, the dynamics of which are dominated by τ1,BR and τ2,BR. This conclusion is supported by similar results from femtosecond IR measurements on SRII,14 HR,13 and PR.17 BR5.12. The transient BR5.12 difference spectra (Figure 9) are dominated by the vibrational features of the BR5.12 chromophore following photoexcitation. The bands due to the residual amount of BR in the sample can be identified spectrally and kinetically. They are used to gauge bands observed in the amide I and amide II regions, showing that the latter are not part of the BR signals but instead part of the BR5.12 fraction. This is described in the following. The IR difference spectra of BR5.12 (Figure 9) exhibit at early delay times the strong ethylenic stretch at 1517 cm-1 and the CdNH stretch at 1633 cm-1. With respect to the corresponding bands in BR, they are shifted by 12 and 7 cm-1, respectively. The instantaneous positive signals at around 1540, 1570, 1610, and 1662 cm-1 are tentatively assigned to S1 vibrational modes of the BR5.12 chromophore or to bands of the BR fraction. As in the fingerprint region, the remaining difference signals at a delay time of 130 ps (Figures 6, 7, and 9) exclusively display bands found in the (pure) K-BR difference spectrum (cp. Figure 5). Especially the bands around 1516, 1529, 1610, and 1640 cm-1 agree well in their spectral position with the difference signals of the ethylenic and the CdNH stretch in BR and K at 40 ps. Likewise, negative bands in the amide I and amide II regions around 1660 and 1550 cm-1 are observed (however, note the discussion on their amplitudes below). In other words, the early difference spectrum (0.15 ps) around the ethylenic stretch and the CdNH stretch is dominated by modes of BR5.12 and merges into a difference spectrum of (pure) BR. No indications are found for residual bands at long delay times that do not belong to BR, except the bands in the amide region. This excludes isomerization or any other photoreaction of the BR5.12 chromophore. Whereas the chromophore bands of the BR fraction in the BR5.12 sample have small amplitudes but are clearly discernible at a long delay time, surprisingly, the bands in the amide I and amide II regions are observed with similar (amide I) or even larger (amide II) amplitude (Figures 6, 7, and 9) and do not

7856

J. Phys. Chem. B, Vol. 113, No. 22, 2009

Figure 10. Absorption transients in the amide I and amide II regions of BR and BB5.12 together with the result of the corresponding global analysis.

vanish on the time scale of our experiment (Figure 10). In (pure) BR, the ratio of the absorbance differences around 1550 (cp. encircled region in Figure 6) and 1529 cm-1 at 40 ps is 1:7, while, in BR5.12, it is about 2:1 (Figure 6). Hence, the signals around 1550 cm-1 in BR5.12 are more than 1 order of magnitude larger for being caused by a BR amount of 15-20%. At 1660 cm-1, an analogous analysis yields somewhat different numbers but the same qualitative result. In (pure) BR, the ratio of the absorbance differences around 1660 (cp. encircled region in Figure 7) and 1640 cm-1 at 40 ps is 1:2, while, in BR5.12, it is about 1:1 (Figure 7). Hence, the signals at 1660 cm-1 in BR5.12 are twice as large for being caused by a BR amount of 15-20%. It is thus concluded that, in (pure) BR5.12, photoexcitation causes changes in amide I and amide II vibrations that persist for much longer than the excited state lifetime of 18 ps and are not a consequence of chromophore isomerization. A global analysis of the BR5.12 (Figure 11) data yields three time constants: τ1 ) 0.5 ( 0.1 ps, τ2 ) 4.7 ( 0.2 ps, and τ3 ) 18.6 ( 0.2 ps. τ1 and τ2 are similar to τ1,BR and τ2,BR and describe processes of the BR fraction. The kinetic development of the ethylenic and CdNH stretch of the electronic ground state in BR5.12 is described by the 18 ps DAS, in accordance with previous results. It shows pronounced negative amplitudes at 1517 and 1633 cm-1 coinciding with the corresponding bleach bands in Figure 9 and indicates the BR5.12 ground state recovery. The positive bands of the 18 ps DAS at 1540 and 1662 cm-1 support the assignment of the corresponding instantaneous difference band to S1 vibrations of BR5.12, disappearing with the excited state decay. The negative band around 1553 cm-1 agrees well with the respective band in the amide II region in (pure) BR. The amide I region of the 18 ps DAS of BR5.12 reveals, as well, negative contributions, superimposed by the positive band at 1662 cm-1. This suggests that the bands in the amide I and amide II regions are subsumed by the fitting procedure with the BR5.12 vibrational dynamics, determined by its S1 decay with 18 ps. Although two processes are apparent, i.e., the S1 decay and the process associated with the nonchromophore bands, the global fit reveals only one time constant instead of two. Two explanations are given. Either the time constant of the second, nonchromophore process is different in BR5.12 and assumes a value closer to 18 ps instead of 11 ps

Gross et al. detected in BR, or the statistical weight of the bands associated with this process is too small compared to the 18 ps kinetics of the dominant vibrational bands of the BR5.12 chromophore. Consequently, the 18 ps DAS combines the S1 decay and the slow part of the nonchromophore process. In Figure 10, absorbance transients at 1556, 1662, and 1669 cm-1 of BR and BR5.12 are depicted as examples for the kinetic characteristics of the bands in the amide II and amide I regions, respectively. In BR, the apparent rise of the bleach signal at 1556 cm-1 occurs with at most a few picoseconds. The positive amplitudes of A1 and A2 at this wavenumber (Figure 8) imply that the risetime is shorter than or equal to 3.2 ps. At 1662 cm-1, the bleach rises significantly faster, dominated by the amplitude of A1 of 0.6 ps. The bleach signals then show a slight absorbance increase determined by a time constant of 11 ps. In BR5.12, the bleach signal at 1556 cm-1 grows within a few picoseconds (